Interactive Engineering, Sydney, NSW, Australia
* Corresponding author. Email: jimbrander@gmail.com (J.B.)
Manuscript submitted March 30, 2023; accepted April 19, 2023; published April 25, 2023.
Abstract: In this article, we introduce an artificial intelligence-based approach to numerical and logical analysis that relies upon a network built up of semantic structures. Our design utilizes an undirected, fully transparent architecture with a capacity for self-extension. It is designed to process information and reason using natural language as its substrate. The network, which we refer to as Active Structure, consists of a semantic network derived from tens of thousands of definitions extracted from a human-readable dictionary. Extensive curation of the dictionary entries was required to translate them into a machine-interpretable form. We outline our solutions to issues that were commonly encountered during the curation process such as circularity, unfounded hierarchical organization, and inadequate detail. We go on to discuss possible applications of Active Structure, including an in-depth example of how it could be employed in the detection of money laundering. An overarching goal of our project is to design a machine intelligence that can overcome the Four Pieces Limit—an upper limit on the capacity of the human mind for conscious processing. We view surpassing this limitation as a crucial step toward the generation of new solutions to a wide variety of problems.
Keywords: Semantic structure, human-readable dictionary, undirected active network, self-extensible, dictionary curation, four pieces limit, anti-money laundering
Cite: Jim Brander and Ryan Sigmundson, "An Active Structure‐Based Artificial General Intelligence,"
Journal of Advances in Artificial Intelligence, Vol. 1, No. 1, pp. 1-19, 2023. doi: 10.18178/JAAI.2023.1.1.1-19
Copyright © 2023 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).
